similar to: Clock Sources under Centos 5.5

I'm wondering under what circumstances the kvm-clock source is exposed to a guest. I've got a bunch of KVM servers running Centos 5.5 and some of the guests have are defaulting to jiffies or apic whilst some are picking up kvm-clock. These guests are a mix of 32bit and 64bit Centos 5.5. Looking at the virtual machine definition file there doesn't appear to be a difference between the

Are their any good guides out there on how to use ksmctl to enable and tune KSM performance on Centos/RedHat At the moment the only guidelines I can find are from the following OLS paper, plus the Linux Kernel Docs. * http://www.kernel.org/doc/ols/2009/ols2009-pages-19-28.pdf * http://www.kernel.org/doc/Documentation/vm/ksm.txt On a basic dual core testbed with 4GB Ram and 4-5 VMs I've

Has anyone here ran performance comparisons between NFS and iSCSI when using network storage for KVM based guests. Also which have people found to be easier for managing live migrates etc. Steve -- Steven Ellis - Bulletin.Net Inc - http://www.bulletin.net

Running Centos 5.4 with KVM on a Dell R610 server and I'd like to control which of the four ethernet interfaces are used for specific tasks My ideal configuration would be eth0 - Host traffic only, no virtual guests. Used for guest mirroring and management. eth1 - NAT guest traffic only, no address for local machine and in some environments in the same zone as eth0 eth2/3 - Allocated to two

Is anyone else having major I/O peaks due to logrotate or other jobs running simultaneously across multiple guests. I have one KVM server running Centos 5.4 with local disk that is seriously suffering as most of the guests rotate their syslog at the same time. Looking at the KVM server I'm seeing 11:00:01 PM CPU %user %nice %system %iowait %steal %idle 03:40:01 AM

[ Ingo, Thomas cc'd in case this issue effects normal jiffies clock too? ] When the Host TSC is unreliable or can change, lguest guests use the jiffies clock. However, the clock value seems to creep upwards in sub-jiffies increments, and then tick_handle_periodic() goes into an infinite loop. Instead, the host writes the current time into the lguest page on every interrupt. This

[ Ingo, Thomas cc'd in case this issue effects normal jiffies clock too? ] When the Host TSC is unreliable or can change, lguest guests use the jiffies clock. However, the clock value seems to creep upwards in sub-jiffies increments, and then tick_handle_periodic() goes into an infinite loop. Instead, the host writes the current time into the lguest page on every interrupt. This

Hello all, I've been digging around on this for a few days and I have not come up with a solution. I have a Compaq ProLiant DL580 (G1, the old tan Compaq) with 4 x 700MHz P-III CPUs and 11GB of memory. I've loaded CentOS 5.5 with Virtualization (Xen) + KVM and patched up to current (full KS packages file list at the end). When it boots, Xen only detects a single CPU as shown in the xm

In a virtualized environment, virtual machines will time share the system with each other and with other processes running on the host system. Therefore, a VM's virtual CPUs (VCPUs) will be executing on the host's physical CPUs (pcpus) for only some portion of time. The VMI exposes a paravirtual view of time to the guest operating systems so that they may operate more effectively in a

In a virtualized environment, virtual machines will time share the system with each other and with other processes running on the host system. Therefore, a VM's virtual CPUs (VCPUs) will be executing on the host's physical CPUs (pcpus) for only some portion of time. The VMI exposes a paravirtual view of time to the guest operating systems so that they may operate more effectively in a

VMI timer code. It works by taking over the local APIC clock when APIC is configured, which requires a couple hooks into the APIC code. The backend timer code could be commonized into the timer infrastructure, but there are some pieces missing (stolen time, in particular), and the exact semantics of when to do accounting for NO_IDLE need to be shared between different hypervisors as well. So

VMI timer code. It works by taking over the local APIC clock when APIC is configured, which requires a couple hooks into the APIC code. The backend timer code could be commonized into the timer infrastructure, but there are some pieces missing (stolen time, in particular), and the exact semantics of when to do accounting for NO_IDLE need to be shared between different hypervisors as well. So

VMI timer code. It works by taking over the local APIC clock when APIC is configured, which requires a couple hooks into the APIC code. The backend timer code could be commonized into the timer infrastructure, but there are some pieces missing (stolen time, in particular), and the exact semantics of when to do accounting for NO_IDLE need to be shared between different hypervisors as well. So

VMI timer code. It works by taking over the local APIC clock when APIC is configured, which requires a couple hooks into the APIC code. The backend timer code could be commonized into the timer infrastructure, but there are some pieces missing (stolen time, in particular), and the exact semantics of when to do accounting for NO_IDLE need to be shared between different hypervisors as well. So

Hi everyone: Just a follow up on the fast clock issue: "Can you double-check that 'cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq' outputs the same number as your 'cpuKHZ' setting in your config.ini file?" There is no cpufreq directory under /sys/devices/system/cpu/cpu0; all I see is "cache" (dir), "crash_notes" (a file) and

Isolate some of the non-VMI timer related changes in Linux. This patch moves the cyc_2_ns conversion code into a common location, eliminating redundant code in hpet and tsc timer implementations, and introduces some macros that may be redefined by the sub-architecture to avoid dependence on APIC routing, CMOS time sync, and testing for broken time hardware (which presumably, does not happen in a

Isolate some of the non-VMI timer related changes in Linux. This patch moves the cyc_2_ns conversion code into a common location, eliminating redundant code in hpet and tsc timer implementations, and introduces some macros that may be redefined by the sub-architecture to avoid dependence on APIC routing, CMOS time sync, and testing for broken time hardware (which presumably, does not happen in a

Convert VMI timer to use clock events, making it properly able to use the NO_HZ infrastructure. On UP systems, with no local APIC, we just continue to route these events through the PIT. On systems with a local APIC, or SMP, we provide a single source interrupt chip which creates the local timer IRQ. It actually gets delivered by the APIC hardware, but we don't want to use the same local

Convert VMI timer to use clock events, making it properly able to use the NO_HZ infrastructure. On UP systems, with no local APIC, we just continue to route these events through the PIT. On systems with a local APIC, or SMP, we provide a single source interrupt chip which creates the local timer IRQ. It actually gets delivered by the APIC hardware, but we don't want to use the same local

Hi, I''m using Xen 3.2.1 with 2.6.26 as Dom0 and DomU in the x86_64 Versions/Builds as provided by Debian Lenny. I noticed that when using "jiffies" as clocksource, the resolution of the timers in a DomU is 4ms, e.g. ping shows only runtimes in 4ms increments [1]. When using "xen" the resolution is fine (didn''t make exact measurements) but xen is a no-go